Impact of referral bias on prognostic studies outcomes: insights from a population-based cohort study on infective endocarditis

Prognostic studies derived from samples of patients managed in tertiary hospitals are subject to referral bias. We aimed characterize this bias using the example infective endocarditis. analyzed data a French population–based cohort, which included 497 with Patients were admitted directly hospital (Group T), non-tertiary and referred NTT) or not NT). compared patients’ characteristics, survival rates prognostic factors between groups. Compared Group T (n = 291), NTT 144) more often males (81.3% vs. 72.5%; P .046), injection drug users (9.7% 4.5%; .033), had frequent surgical indications (78.5% 64.3%; .003). NT 62), 67.7%; .034) 19.4%; < .001). One-year was higher + than (73.0% 56.1%; .01). hazard ratios estimates varied across When mixing those hospitals, validity characteristics description, estimates, is threatened by